sign in sign up
<<
Home , Baroclinity, Kuroshio Current, Typhoon Morakot

remote sensing

Article Satellite Observations of -Induced Surface Variability in the Region off Northeastern

Yi-Chun Kuo 1,2, Ming-An Lee 2,3,* and Yi Chang 4

1 Institute of , National Taiwan University , Taipei 106, Taiwan; [email protected] 2 Department of Environmental Biology and Fisheries Science, National Taiwan University, 20224, Taiwan 3 Center of Excellence for , National Taiwan Ocean University, Keelung 20224, Taiwan 4 Graduate Institute of Marine Affairs, National Sun Yat-sen University, City 80424, Taiwan; [email protected] * Correspondence: [email protected]

 Received: 11 September 2020; Accepted: 10 October 2020; Published: 12 October 2020 

Abstract: Typhoon-induced cooling in the cold dome region off northeastern Taiwan has a major influence on ocean biogeochemistry. It has previously been studied using numerical models and hydrographic observations. Strong cooling is related to upwelling of the Kuroshio subsurface water accompanied by the westward intrusion of the by Kuroshio water. By employing satellite observations, local measurements, and a reanalysis of model data, this study compared 18 typhoon-induced (SST) responses in the cold dome region and determined that SST responses can differ dramatically depending on the relative location of a typhoon path, the Kuroshio Current, and the topography off northeastern Taiwan. The results indicated that local westward and northward stress is positively correlated with upwelling intensity. Decreased northward transport in the created a condition that favored the Kuroshio intrusion, thus, the typhoon-induced change in Taiwan Strait transport was also positively correlated with the intensity of cooling. However, the strength of Ekman pumping was weakly correlated with the intensity of SST cooling. Nevertheless, Ekman pumping helped reduce the cover of warm water, facilitating the intrusion of the Kuroshio Current.

Keywords: sea surface temperature; upwelling; typhoon; Kuroshio

1. Introduction The Kuroshio Current flows northward along the east coast of Taiwan, with its principal stream turning northeast along the continental shelf of the East Sea (ECS) after leaving the coast (Figure1). Hydrographic observations and current measurements have confirmed the presence of subsurface Kuroshio waters off northeastern Taiwan [1]. Persistent upwelling off this area is associated with weak local cyclonic circulation and the intrusion of Kuroshio waters interacting with the ocean topography [2]. The upwelling brings cold and saline Kuroshio waters to the ECS continental shelf, forming a cold dome region. Higher concentrations of nutrients can be found in the upwelled waters, thus, the cold dome is a major fishing ground and its physical properties, along with its spatial and temporal variations, have been widely studied [1–9].

Remote Sens. 2020, 12, 3321; doi:10.3390/rs12203321 www.mdpi.com/journal/remotesensing Remote Sens.Remote2020 Sens., 12 ,2020 3321, 12, x FOR PEER REVIEW 2 of 16 2 of 14

Figure 1. around Taiwan. The blue circle in the zoom-in plot denotes the cold dome region. Figure 1. Bathymetry around Taiwan. The blue circle in the zoom-in plot denotes the cold dome Shen etregion. al. [3] attributed the summertime Kuroshio intrusion to a shoreward force, similar to the centrifugal forceShen et caused al. [3] attributed by the Kuroshio the summertime as it follows Kuroshio (approximately) intrusion to a shoreward the curvature force, similar of theto the isobaths off northeasterncentrifugal Taiwan. force caused The by cold the Kuroshio dome is as modulated it follows (approximately) by local current the curvature circulation. of the Toisobaths the west of Taiwan,off the northeastern Taiwan Strait Taiwan. (TS)’s The warm cold currentsdome is modulated flow northward by local to current join the circulation. ECS. TS outflowTo the west is relatedof to the strengthTaiwan, of the Taiwan cold dome. Strait (TS)’s Stronger warm (weaker)currents flow TS northward currents to often join the accompany ECS. TS outflow a weaker is related (stronger) Kuroshioto intrusionthe strength because of the cold of dome. changes Stronger in the (weaker) TS currents field. Theoften circulationaccompany a oweakerff northern (stronger) Taiwan is Kuroshio intrusion because of changes in the pressure field. The circulation off northern Taiwan is also stronglyalso strongly modulated modulated by East by East Asian Asian . monsoons. The Kuroshio Kuroshio current current east eastof Taiwan of Taiwan is farther is farther offshoreoffshore during during summer, summer, and and thus, thus, intrusion intrusion to to thethe ECS continental continental shelf shelf is weaker is weaker than in than winter. in winter. The mechanismThe mechanism of the of upwelling the upwelling is complicated, is complicated, which which means the the temporal temporal and andspatial spatial scales scalesof the of the cold dome vary. The radius of the cold dome is approximately 50 km, and variations in the shape of cooling area are large [2]. frequently occur over the tropical western Pacific and pass near the southern ECS. They can trigger blooms and increase by inducing vertical mixing and upwelling near their paths and around Taiwan [10,11]. In an open ocean, a typhoon can induce significant upper ocean cooling, the primary process of which depends on the ratio of the typhoon’s moving speed to its radius. One major process is Ekman pumping induced by wind stress Remote Sens. 2020, 12, 3321 3 of 14 curl, and another is the vertical mixing or entrainment that accompanies typhoon-induced inertial oscillations [12,13]. However, the presence of the solid boundary, shoaling topography, and the coastal current in the continental shelf region means that a typhoon-induced ocean response is more complicated and unique. Studies have observed significant typhoon-enhanced cooling (4–8 ◦C) [9,10,14] in the cold dome region. Using hydrographic data, Tsai et al. [14] concluded that typhoon-induced cooling in the cold dome region is due to the Kuroshio being driven by strong typhoon onto the continental shelf north of Taiwan. Morimoto et al. [15] further investigated the fluctuations and mechanism of Kuroshio axis movement using sea surface current data. They proposed that the continuous blowing of southerly winds may push the Kuroshio axis onto the shelf and persist for a period of time after a typhoon has passed. Chang et al. [10] suggested that a drop in sea surface temperature (SST) might be caused by the Ekman pumping induced by typhoons. Although the mechanism of enhanced cooling associated with Kuroshio upwelling in the southern ECS has been investigated in numerous studies, these have primarily been case studies. Moreover, the enhanced cooling in the cold dome region that is associated with a typhoon is not limited to a particular type of track [14]. A more comprehensive study on typhoon-induced cooling off northeastern Taiwan, one that determines the different intensities and tracks of a typhoon, would be helpful in predicting typhoon-induced physical and biological responses in the cold dome region. To achieve this, high-resolution SST monitoring images are required to identify the pattern of SST changes associated with a typhoon. This study, therefore, investigated enhanced cooling off the northeastern coast of Taiwan after the passing of various typhoons. We used Advanced Very High-Resolution Radiometer (AVHRR) SST data to identify areas of cooling. However, clear daily images were unavailable due to cover, therefore, one-week composite images were used. Upwelling intensity was determined according to the SST difference between the mean location (25.25–25.75◦N, 122–122.5◦E) of the cold dome and the Kuroshio region southeast of the dome (24.5–25.0◦N, 122.0–122.5◦E) off eastern Taiwan. Higher contrast between the SST of the cold dome and the surrounding Kuroshio region was associated with stronger upwelling in the cold dome. We found that the strength and duration of southeasterly wind stress and the change in water transport in the TS were significantly correlated with upwelling intensity. The remainder of this paper is organized as follows. Section2 presents the data sources and the index calculation of factors, Section3 presents the results, and Section4 presents the discussion and conclusions.

2. Methods AVHRR SST level 2 data of cloud and land masking from June 2009 to August 2013 were collected from series National Oceanic and Atmospheric Administration satellite (NOAA 16~18) produced by regional High Resolution Picture Transmission (HRPT) data library at National Taiwan Ocean University. Navigation and cloud-detection techniques developed by Sakaida and Kawamura [16] were then employed. The multichannel SST (MCSST) algorithm [17] was used to produce pass SST data at a spatial resolution of 1.1 km, all passes in both daytime and nighttime were then composited into daily average images. Lee et al. [18] validated the SST algorithm by comparing 1998–2002 AVHRR-based MCSSTs with in situ data. The MCSSTs exhibited a small bias of 0.009 ◦C, with Root Mean Square Deviation (RMSD) of 0.64 ◦C. Weekly composite SST data were generated by arithmetically averaging all the available scenes in each month on a pixel-by-pixel basis (excluding missing data and ). The ocean color level 2 products with 1.1 km resolution of MODIS (Moderate Resolution Imaging Spectroradiometer) onboard both Aqua and Terra satellites were collected from Ocean Color Web (https://oceancolor.gsfc.nasa.gov/) of NASA (National Aeronautics and Space Administration). Because the activated coverage was low, weekly mean maps are composed before and after the typhoon passage. The best track data were obtained from the Digital Typhoon website (http://agora.ex.nii.ac.jp/ digital-typhoon/) operated by ’s National Institute of Informatics. These provided access to SST Remote Sens. 2020, 12, x FOR PEER REVIEW 4 of 16 collected from Ocean Color Web (https://oceancolor.gsfc.nasa.gov/) of NASA (National Aeronautics and Space Administration). Because the activated coverage was low, weekly mean maps are composed before and after the typhoon passage. Remote Sens. 2020, 12, 3321 4 of 14 The best track data were obtained from the Digital Typhoon website (http://agora.ex.nii.ac.jp/digital-typhoon/) operated by Japan’s National Institute of Informatics. imagesThese provided of 18 typhoons access thatto SST passed images through of 18 typhoons the study areathat passed from 2009 through to 2013. the Weekly study area mean from maps 2009 were to composed2013. Weekly before mean and maps after were each composed typhoon had before passed. and after each typhoon had passed. The 18 typhoon tracks are plotted in Figure 22.. TheThe typhoon season lasts from May to September. Most of the 18 typhoon tracks headed northwest andand came from the east of Taiwan. Three typhoons headed northeast, northeast, passing passing east east of of Taiwan. Taiwan. Most Most typhoons typhoons passed passed north north to the tothe cold cold dome dome region, region, six sixtyphoons typhoons passed passed south south to the to thecold cold dome dome region region,, and and four four typhoons typhoons passed passed east east to Taiwan. to Taiwan.

Figure 2. Tracks of the 18 typhoons that passed through the vicinityvicinity ofof Taiwan.Taiwan. The time interval between each markermarker isis oneone day.day. TheThe numbernumber beside beside each each marker marker represents represents the the “day.” “day.” The The “month” “month” is omittedis omitted because because this this is obviousis obviou froms from the the life life of of the the typhoon. typhoon. P The cooling area is defined as Ac = A , where A is the unit of area (1 1 km) in which ∆SST > 1°C, The cooling area is defined as Ac = ∑Ai i, where Ai i is the unit of area× (1 × 1 km) in which ∆SST > where1, where∆SST ∆SSTis the is SST the di SSTfference difference derived derived from thefrom prestorm the prestorm one-week one-week prestorm prestorm composite composite image compared with the one-week poststorm image. Upwelling intensity is defined as Ic = T Tc, where T image compared with the one-week poststorm image. Upwelling intensity is definedk −as Ic = Tk − Tck, is the average SST in the cold dome region (24.5–25.0 N, 122.0–122.5 E; Figure1), which represents where Tk is the average SST in the cold dome region◦ (24.5–25.0°N, 122.0–122.5°E;◦ Figure 1), which the temperature in the Kuroshio region off northeastern Taiwan. T is the average SST in a box represents the temperature in the Kuroshio region off northeastern Taiwan.c Tc is the average SST in regiona box region (25.25–25.75 (25.25–25.75°N,◦N, 122.0–122.5 122.0–122.5°E),◦E), which representswhich represents the temperature the temperature in the cold in domethe cold region dome off northeasternregion off northeastern Taiwan. Higher Taiwan. contrast Higher between contra the SSTst between of the cold the dome SST andof itsthe surrounding cold dome Kuroshio and its region is related to stronger upwelling in the former. To estimate the potential influence of upwelling surrounding Kuroshio region is related to strongerP upwelling in the former. To estimate the potential associated with total heat loss, we defined Sc = ∆ SST A . influence of upwelling associated with total heat loss, wei × definedi Sc = ∑ ∆ SST i × Ai. The factorsfactors associatedassociated with with upwelling upwelling intensity intensit arey windare wind speed, speed, wind wind stress curlstress (inducing curl (inducing Ekman pumping),Ekman pumping), the TS current,the TS current, and Kuroshio and Kuroshio velocity. velo Thecity. existence The existence of various of various typesof types typhoon of typhoon tracks meanstracks means that the that direction the direction of major of wind major forcing wind in fo thercing study in the region study varied region among varied cases. among To synthesize cases. To wind forcing, wind stress was decomposed into U+ (eastward direction), U (westward direction), synthesize wind forcing, wind stress was decomposed into U+ (eastward direction),− U− (westward V+ (northward direction), and V (southward direction) components. We used hourly wind data from − Taiwan’s Central Bureau to investigate variations in wind speed during typhoon periods in the Pengjiayu Islands (25◦37’31”N 122◦04’32”E) and over the upwelling region. A typical time scale Remote Sens. 2020, 12, x FOR PEER REVIEW 5 of 16

Remotedirection), Sens. 2020 V+, 12 (northward, 3321 direction), and V− (southward direction) components. We used hourly5 of 14 wind data from Taiwan’s to investigate variations in wind speed during typhoon periods in the Pengjiayu Islands (25°37’31”N 122°04’32”E) and over the upwelling region. forA typical a storm time moving scale across for a storm Taiwan moving or its vicinityacross Taiw is approximatelyan or its vicinity three is days.approximatelyMorimoto three et al. days. [13] alsoMorimoto observed et anal. approximate[13] also observed three-day an approximate lag between windthree-day and maximumlag between cooling. wind Weand thus maximum chose acooling. three-day We average thus chose wind a three-day speed during average the typhoonwind speed forcing during period, the typhoon with average forcing wind period, forcing with definedaverage bywind an hourlyforcing wind defined speed by largeran hourly than wind 10 m /s.speed The larger wind fieldthan was10 m/s. obtained The wind from field blended was seaobtained wind datafrom produced blended sea by NOAAwind data (http: produced//www.ncdc.noaa.gov by NOAA (http://www.ncdc.noaa.gov/oa/rsad/air-/oa/rsad/air-sea/seawinds.html#data). Tosea/seawinds.html#data). measure the Ekman pumping To measure velocity, the theEkman wind pumping stress vector velocity, T was the calculatedwind stress using vector the T bulk was τ = C U U C transfercalculated formula using theρ abudlk 10transfer10, where formulad is τ the = drag coeffi¸ where cient. For moderateis the drag and coefficient. strong wind, For themoderate following and observations strong wind, showed the following that the observations coefficient increased showed that with the the coefficient increase ofincreased the wind with speed the andincrease the linear of the relation wind speed was established and the linear between relation the dragwas established coefficient and between the mean the drag wind coefficient speed [19– and22] (e.g.,the mean Large wind and Pondspeed, 1981;[19–22] Guan (e.g., and Large Xie, and 2004; Pond, Smith, 198 1980;1; Guan Wu, and 1980). Xie, In2004; the Smith, present 1980; study, Wu, we 1980). use theIn the Cd present determined study, by we Large use and the PondCd determined [19] such thatby Large and Pond [19] such that

( 1.14 × 103, < 101 1.14 10− , U10 < 10ms− C = = × (1)(1) d (0.49(0.49+ +0.065 0.065U )) 10×103, ,U ≥10 10ms 1 10 × − 10 ≥ − The local inertial period was approximately 26 hours. We thus used one-day mean wind speed The local inertial period was approximately 26 hours. We thus used one-day mean wind speed to calculate maximum daily local Ekman pumping: E= (∇× ). To evaluate the change in water to calculate maximum daily local Ekman pumping: E = 1 →τ . To evaluate the change in ρf ∇ × watertransport transport in the inTS the and TS the and Kuroshio the Kuroshio region region off eastern off eastern Taiwan, Taiwan, we used we data used from data a from daily a daily1/12°- resolution global produced by the Hybrid Coordinate Ocean Model (HYCOM; 1/12◦-resolution global ocean reanalysis produced by the Hybrid Coordinate Ocean Model (HYCOM; https://www.hycom.org/hycom). Transport in the upper 10 m and along a 25°N transect in the TS https://www.hycom.org/hycom). Transport in the upper 10 m and along a 25◦N transect in the TS was chosenwas chosen to represent to represent TS outflow. TS outflow. For the For Kuroshio the Ku intrusion,roshio intrusion, westward westward transport transport along the along transect the transect 122.0–122.25°E, 25°N (0–10 m in depth) was calculated. Northward Kuroshio transport along 122.0–122.25◦E, 25◦N (0–10 m in depth) was calculated. Northward Kuroshio transport along the the transect 122.0–122.25°E, 24.5°N (0–10 m in depth) was chosen to represent the Kuroshio velocity transect 122.0–122.25◦E, 24.5◦N (0–10 m in depth) was chosen to represent the Kuroshio velocity off easternoff eastern Taiwan. Taiwan. For eachFor each current current transport, transport, the three-day the three-day average average before before and after and theafter encounter the encounter with eachwith typhoon each typhoon was calculated. was calculated.

3.3. ResultsResults MonthlyMonthly climatological climatological SSTs SSTs are are shown shown in in Figure Figure3. 3. The The SST SST gradient gradient in in the the study study area area is is weaker weaker inin summer summer than than in winter.in winter. SSTs SSTs ranged ranged from from 24 ◦C 24 (May) °C to(May) 26.5 ◦toC (August).26.5 °C (August). The cold domeThe cold appeared dome inappeared monthly in SST monthly images SST from images May from to September. May to September. The circular The cold circular region cold is lessregion distinct is less in distinct October. in ThisOctober. is due This to a is larger due to portion a larger of portion the Kuroshio of the Ku flowingroshio onto flowing the continentalonto the continental shelf, weakening shelf, weakening the SST frontthe SST [2]. front [2].

FigureFigure 3. 3.Monthly Monthly climatologicalclimatological seasea surfacesurface temperaturestemperatures ( ◦(°C,C, averagedaveraged from from 1993 1993 to to2013) 2013)from from May May toto October. October.

Remote Sens. 2020, 12, x FOR PEER REVIEW 6 of 16 Remote Sens. 2020, 12, 3321 6 of 14 Figure 4 depicts the difference between prestorm and poststorm SSTs (one-week composite images) for the 18 typhoons. Most typhoons induced significant cooling off northeastern Taiwan at a Figure4 depicts the di fference between prestorm and poststorm SSTs (one-week composite images) magnitude of approximately 2–4 °C. However, five cases of warming also occurred (Typhoons 1, 9, for the 18 typhoons. Most typhoons induced significant cooling off northeastern Taiwan at a magnitude 11, 13, and 18). Notably, significant warming of up to 3 °C appeared in the study region for Typhoon of approximately 2–4 ◦C. However, five cases of warming also occurred (Typhoons 1, 9, 11, 13, and 18). 13. This raises the question of whether SST changes after typhoons are part of normal seasonal Notably, significant warming of up to 3 ◦C appeared in the study region for Typhoon 13. This raises variations in the region. To clarify these warming anomalies, the SST differences from 10-year the question of whether SST changes after typhoons are part of normal seasonal variations in the climatological means are depicted (Figure 5). These indicate that warming still occurred in these five region. To clarify these warming anomalies, the SST differences from 10-year climatological means are cases. The SST standard deviation from the climatological mean exceeded 1.5 °C in Typhoons 9 and depicted (Figure5). These indicate that warming still occurred in these five cases. The SST standard 13. We concluded that the warming phenomena off northeastern Taiwan are associated with typhoon deviation from the climatological mean exceeded 1.5 ◦C in Typhoons 9 and 13. We concluded that the events. warming phenomena off northeastern Taiwan are associated with typhoon events.

Figure 4. Difference between poststorm and prestorm sea surface ( C, one-week composite Remote Sens. 2020, 12, x FOR PEER REVIEW ◦ 7 of 16 images) for the 18 typhoons shown in Figure2. Figure 4. Difference between poststorm and prestorm sea surface temperatures (°C, one-week composite images) for the 18 typhoons shown in Figure 2.

Figure 5. Difference between poststorm (one-week composite images) and monthly climatological sea Figuresurface 5. temperatures Difference between (◦C) for poststorm Typhoons (one-week 1, 9, 11, 13, compos and 18.ite images) and monthly climatological sea surface temperatures (°C) for Typhoons 1, 9, 11, 13, and 18.

Figure 6 displays climatological mean (2009–2013) of monthly Chl-a concentration and the prestorm Chl-a concentration for the 18 typhoons. The variation of the Chl-a spatial pattern during May–September is not obvious. The Chl-a generally decreased off shore. The prestorm Chl-a concentration in the cold dome area is higher than the climatological mean in Typhoon 2, 5, 6, 7, and 17 cases. This is due to the shorter duration between the previous and present typhoons. The differences between prestorm and poststorm Chl-a concentration are shown in Figure 7. The weekly maps of Chl-a concentration revealed that Chl-a generally increased between 0.5 and 2 mg/m3. The strongest increase occurred in Typhoon 3. The high Chl-a patches match with the sea surface cooling in terms of location (Figures 4 and 7). The colder and nutrient-rich water in the coastal region of Taiwan is probably originated from the Kuroshio subsurface water. The increased Chl-a anomaly is most likely a phytoplankton bloom due to the new nutrients supplied by the uplifted deep water, this phenomenon is supported by field observations [23]. In some cases, the difference of Chl-a concentration is negative, i.e., Typhoon 5, 6, 7, 14 and 17. Most of their prestorm states have higher Chl-a concentration than the climatological mean, except in Typhoon 14. The typhoon-induced turbulent mixing decreased the local concentration of Chl-a. Note that for some sea surface warming cases (Typhoons 1 and 18), the Chl-a concentration still increased. This indicates that besides upwelling, the source of the higher Chl-a concentration could come from the China coastal region under the typhoon wind.

Remote Sens. 2020, 12, 3321 7 of 14

Figure6 displays climatological mean (2009–2013) of monthly Chl-a concentration and the prestorm Chl-a concentration for the 18 typhoons. The variation of the Chl-a spatial pattern during May–September is not obvious. The Chl-a generally decreased off shore. The prestorm Chl-a concentration in the cold dome area is higher than the climatological mean in Typhoon 2, 5, 6, 7, and 17 cases. This is due to the shorter duration between the previous and present typhoons. The differences between prestorm and poststorm Chl-a concentration are shown in Figure7. The weekly maps of Chl-a concentration revealed that Chl-a generally increased between 0.5 and 2 mg/m3. The strongest increase occurred in Typhoon 3. The high Chl-a patches match with the sea surface cooling in terms of location (Figures4 and7). The colder and nutrient-rich water in the coastal region of Taiwan is probably originated from the Kuroshio subsurface water. The increased Chl-a anomaly is most likely a phytoplankton bloom due to the new nutrients supplied by the uplifted deep water, this phenomenon is supported by field observations [23]. In some cases, the difference of Chl-a concentration is negative, i.e., Typhoon 5, 6, 7, 14 and 17. Most of their prestorm states have higher Chl-a concentration than the climatological mean, except in Typhoon 14. The typhoon-induced turbulent mixing decreased the local concentration of Chl-a. Note that for some sea surface warming cases (Typhoons 1 and 18), the Chl-a concentration still increased. This indicates that besides upwelling, the source of the higher Chl-a Remoteconcentration Sens. 2020, could12, x FOR come PEER from REVIEW the China coastal region under the typhoon wind. 8 of 16

Figure 6. Climatological mean (2009–2013) of monthly Chl-a concentration (mg/m3, May–Sep) and Figure 6. Climatological mean (2009–2013) of monthly Chl-a concentration (mg/m3, May–Sep) and the the prestorm Chl-a concentration (mg/m3, one-week composite images) for the 18 typhoons shown in prestorm Chl-a concentration (mg/m3, one-week composite images) for the 18 typhoons shown in Figure2. Figure 2.

Remote RemoteSens. 2020 Sens., 202012, x, 12FOR, 3321 PEER REVIEW 8 of 14 9 of 16

Figure 7. Difference between poststorm and prestorm Chl-a concentration (mg/m3, one-week composite Figure 7. Difference between poststorm and prestorm Chl-a concentration (mg/m3, one-week images) for the 18 typhoons shown in Figure2. composite images) for the 18 typhoons shown in Figure 2. Weekly merged altimetry data obtained from the AVISO database (http://aviso.altimetry.fr) show Weeklythe corresponding merged altimetry changes in data sea surface obtained height from (SSH; th Figuree AVISO8). Ocean database surface (http: cooling//aviso.altimetry.fr) (warming) show wasthe relatedcorresponding to a decreased changes (increased) in sea SSH surface caused byheight hydrostatic (SSH; equilibrium. Figure 8). This Ocean indicates surface that acooling (warming)fall in SSTwas is related accompanied to a bydecreased a cooler water (increased) column underneath,SSH caused which by resultshydrostatic from the equilibrium. upwelling This phenomenon. The SSH anomaly (SSHA) also indicates a longer time scale for the cooling state, and the indicates that a fall in SST is accompanied by a cooler underneath, which results from shapes of the SSHAs are different from those of SSTs. The derived from SSH the upwelling(retrieved fromphenomenon. https://www.aviso.altimetry.fr The SSH anomaly/en/data.html (SSHA)) also is also indicates shown in Figurea longer8. The time spreading scale for the coolingof state, cooler and (warmer) the shapes surface of water the hadSSHAs not reached are diff aerent geostrophic from those balance, of andSSTs. therefore, The geostrophic surface flow current derivedvaried from with SSH time. (retrieved In addition, from the resolution https://www.aviso.altimetry.fr/en/ of SSH and SST images also diffdata.html)ers. is also shown in Figure 8. TheTable spreading1 presents theof cooler correlations (warmer) between surface upwelling water intensity had not (I creached), the cooling a geostrophic area (Ac), and balance, the and therefore,wind surface and current flow factors. varied These with show time. that east–west In addi windtion, speedthe resolution and Kuroshio of intrusion SSH and velocity SST during images also the forced period significantly correlate with the upwelling intensity of the cold dome. TS transport, differs. wind speed, and wind duration significantly correlate with the cooling area. The relationship between the intensity of cooling and the change in current field is discussed in Section3.

Remote Sens. 2020, 12, 3321 9 of 14 Remote Sens. 2020, 12, x FOR PEER REVIEW 10 of 16

Figure 8. Difference between poststorm and prestorm weekly merged sea surface height and geostrophic currentFigure vectors8. Difference for the 18between typhoons poststorm shown in and Figure prestorm2. The vectors weekly are merged represented sea bysurface the di heightfference and of geostrophicgeostrophic current current in vectors zonal andfor the meridional 18 typhoons components. shown in Figure 2. The vectors are represented by the difference of geostrophic current in zonal and meridional components.

Table 1. Correlation coefficients between upwelling intensity (Ic), the cooling area (Ac), and the wind Table 1 presents the correlations between upwelling intensity (Ic), the cooling area (Ac), and the and current factors. Values reaching a 90% significance level are marked with ‘*’. wind and current factors. These show that east–west wind speed and Kuroshio intrusion velocity during the forced period significantly correlate with the upwellingStrong intensity Wind of the cold dome. TS U Wind (m/s) V Wind (m/s) transport, wind speed, and wind duration significantly correlateDuration (with>10 m /thes) cooling area. The ∆SST (Ic) 0.72 * 0.40 * 0.22 relationship between the intensity− of cooling and the change in current field is discussed in Section 3. Cooling Area (Ac) 0.49 * 0.41 * 0.40 * − − Maximum Daily Ekman TS Northward Kuroshio Kuroshio Intrusion 2 Pumping (cm/s) Transport 1 Transport 3

∆SST (Ic) 0.06 0.27 0.43 * 0.22 − − − Cooling Area (Ac) 0.06 0.59 * 0.59 0.11 1 2 Northward water transport in the upper 50 m and along a 25◦N transect. Westward transport in the upper 3 50 m and along the transect 122.0–122.5◦E, 25◦N. Northward Kuroshio transport along the transect 112.0–112.5◦E, 24.5◦N (0–50 m in depth).

3.1. Change in Current Field The Kuroshio intrusion anomaly was positively correlated with upwelling intensity, which is represented by the difference between SSTs in the cold dome region and the Kuroshio region to the east of Taiwan (Table1). When the intrusion was stronger, the upwelling intensity increased. This result is consistent with previous research [12] that concluded that the Kuroshio intrusion is the primary cause of enhanced cooling in the cold dome area. This situation was reversed in warming cases (Typhoons 1, 9, 11, 13, and 18), where the eastward flow anomaly in the region off northeastern Taiwan

Remote Sens. 2020, 12, 3321 10 of 14 was associated with the warming event. The variation in TS transport correlated highly with the cooling area (Ac). Larger typhoon-induced southward TS transport was related to a broader cooling region off northeastern Taiwan. Furthermore, the upwelling intensity was negatively correlated with flow transport in the TS, hence, decreased (increased) TS flow during the forced period corresponded to a positive (negative) intensity (nonsignificant) and area (significant). The variations in TS transport were negatively correlated with the Kuroshio intrusion (correlation coefficient of 0.59). This result indicates that the often-stronger Kuroshio intruding onto the ECS continental shelf was accompanied by a smaller TS northward flow. Chern and Wang (1992) proposed that the TS outflow controls the intensity of the Kuroshio intrusion, with a stronger TS flow during the summer. The relationship between the intrusion of the Kuroshio and the retreat of the TS current is consistent with the hydrographic observations and current measurements of Tang et al. [1]. Figure8 does not display the opposite direction of the geostrophic velocity anomaly between the Kuroshio east of Taiwan and the TS current. Modification of the circulation pattern occurred during the typhoon forced period, with the Kuroshio intruding onto the ECS continental shelf off northeastern Taiwan. During the poststorm period, the Kuroshio began to return to its normal status. Morimoto et al. [15] suggested that the Kuroshio intrusion is aided by a stronger typhoon-induced Kuroshio Current to the east of Taiwan. There is a positive correlation, albeit weak, between upwelling intensity and the northward Kuroshio velocity east of Taiwan (122.0–122.4◦E, 24◦N). Figure9a,b shows the linear relationship between the change in TS transport and S c (defined in Section2) and the Kuroshio intrusion and S c, respectively. The shape of the cold SST in the cold dome region changed from the typhoon forced period to the poststorm period, and therefore, the Sc calculated from the composite image might only be approximate. Nevertheless, the R2 in Figure9a,b was 0.46 andRemote 0.44, Sens. respectively. 2020, 12, x FOR PEER REVIEW 12 of 16

Figure 9. LinearFigure relationship9. Linear relationship between between (a )(a the) the changechange in inTaiwan Taiwan Strait Straittransport, transport, (b) the Kuroshio (b) the Kuroshio intrusion, (c) east–west wind stress, (d) the Ekman pumping velocity, and Sc (×104 km2 °C) 4(defined2 in intrusion, (c) east–west wind stress, (d) the Ekman pumping velocity, and Sc ( 10 km ◦C) (defined in Section 2), respectively. Numbers correspond to the typhoons listed in Figure 2. The× red dash-lines Section2), respectively. Numbers correspond to the typhoons listed in Figure2. The red dash-lines show the 95% confidence interval. show the 95% confidence interval. 3.2. Relationship between Cooling and Typhoon Winds The U-wind correlated highly with the upwelling intensity (a correlation coefficient of 0.72). This indicates that stronger westward wind stress was related to stronger ocean surface cooling in the cold dome region and was caused by the Kuroshio intrusion. In addition, ocean surface warming accompanied larger eastward wind stress during the storm forced period. Figure 9c,d shows the linear relationship between east–west wind stress and Sc and the Ekman pumping velocity and S, respectively. The R2 in Figure 9c,d reached 0.33 and 0.02, respectively. The north–south wind stress did not display a significant correlation with the upwelling intensity (a correlation coefficient of −0.4). In general, southerly wind stress might relate to stronger ocean surface cooling in the cold dome region. Southerly winds thus strengthen the in the Kuroshio region off eastern Taiwan by inducing the eastward Ekman drift. Enhanced northward transport of the Kuroshio to the east of Taiwan favors the intrusion process, however, southerly winds might not be a critical factor for cooling in the cold dome region. The relationship between Ac and wind stress was similar to that of Ic and wind stress, although the correlation was less significant. Overall, strong easterly winds played a major role in the intrusion of subsurface Kuroshio water onto the ECS continental shelf. During the typhoon forced period, the often rotated between the front half and rear half of a typhoon on a westward track, the wind-induced would then change accordingly. However, the rotating wind stress continued to inject

Remote Sens. 2020, 12, 3321 11 of 14

3.2. Relationship between Cooling and Typhoon Winds The U-wind correlated highly with the upwelling intensity (a correlation coefficient of 0.72). This indicates that stronger westward wind stress was related to stronger ocean surface cooling in the cold dome region and was caused by the Kuroshio intrusion. In addition, ocean surface warming accompanied larger eastward wind stress during the storm forced period. Figure9c,d shows the linear relationship between east–west wind stress and Sc and the Ekman pumping velocity and S, respectively. The R2 in Figure9c,d reached 0.33 and 0.02, respectively. The north–south wind stress did not display a significant correlation with the upwelling intensity (a correlation coefficient of 0.4). In general, southerly wind stress might relate to stronger ocean − surface cooling in the cold dome region. Southerly winds thus strengthen the pressure gradient in the Kuroshio region off eastern Taiwan by inducing the eastward Ekman drift. Enhanced northward transport of the Kuroshio to the east of Taiwan favors the intrusion process, however, southerly winds might not be a critical factor for cooling in the cold dome region. The relationship between Ac and wind stress was similar to that of Ic and wind stress, although the correlation was less significant. Overall, strong easterly winds played a major role in the intrusion of subsurface Kuroshio water onto the ECS continental shelf. During the typhoon forced period, the wind direction often rotated between the front half and rear half of a typhoon on a westward track, the wind-induced ocean current would then change accordingly. However, the rotating wind stress continued to inject energy into the ocean. This energy flux is most efficient for a wind–current resonant case [24]. Wind-induced turbulent mixing continued during the strong-wind period. We calculated the strong-wind (daily mean >10 m/s) duration for each typhoon case. The result indicated a nonsignificant correlation with the Ic and Ac. We, therefore, infer that local wind-induced mixing did not exert a dominant effect on the SST response to the typhoon. Wind stress curl, related to Ekman pumping, did not correlate highly with the ocean surface cooling. The numerical model developed by Tsai et al. [25] showed a negative wind curl associated with downward Ekman pumping velocity occurring off northeastern Taiwan during the entire Morakot forced period, however, cooling in the cold dome region was significant.

4. Discussion Significant SST variations in the cold dome region were found to be associated with typhoons passing through the vicinity of Taiwan. The location and direction of the storm tracks were, however, very different. The difference in SST between the cold dome area and the Kuroshio region (Ic) was chosen to represent the cooling induced by the upwelling associated with the Kuroshio intrusion. The result suggests that the path of a typhoon might be associated with an enhanced or weakened intrusion of the Kuroshio Current off northeastern Taiwan. We found that the east–west wind stress of the storm was significantly correlated with the difference of Ic from the poststorm to prestorm period. Ocean surface cooling tended to be associated with larger average easterly winds, such as those in Typhoons 3, 5, 10, 12, 15, and 16, which were able to drive the Kuroshio intrusion onto the shelf. In these cases, the path went westward or northwestward through Taiwan. The strong easterly winds over Taiwan were accompanied by northerly (or northeasterly) winds over the TS. The reduced TS outflow created a condition that promoted the Kuroshio intrusion. In the cases of Typhoons 3, 12, and 16, the northerly wind in the front half of the typhoon during the forced period caused a strong southward flow anomaly in the TS. This caused a smaller pressure gradient on the western side of the cold dome area and promoted the expansion of the cold water region. Conversely, Typhoons 1, 4, 9, 13, 17, and 18, which decreased Ic, tended to be associated with larger westerly winds. The association between increasing SST in the cold dome region and typhoons had not previously been studied because typhoon-induced ocean cooling is more frequent and can affect the ecosystem through biochemical responses associated with upwelling. However, warming is an intriguing phenomenon because it is an abnormal typhoon-induced ocean response. Moreover, increased SSTs that accompany an increased SSHA might cause a different air–sea interaction. The ocean response to Typhoon Haikui exhibited the strongest warming and SSHA change. We thus chose the Haikui case to Remote Sens. 2020, 12, 3321 12 of 14 examine a possible variation in ocean circulation, causing ocean surface warming events. The Morakot case was chosen to represent a typical drop in typhoon-induced SST in the cold dome area. Figure 10a shows the daily surface temperature and velocity deviating from the day before Typhoon Morakot (06/08/2009). Daily mean model data were provided by the HYCOM GLBa0.08 experiment. The typhoon caused a strong southward current during the front half of the wind field (August 7 to 8). A significant drop in SST occurred in the cold dome area, accompanied by an enhanced Kuroshio intrusion. The flow anomaly turned northward behind the rear half of the typhoon. The cold waterRemote created Sens. 2020 by, 12 the, x FOR Kuroshio PEER REVIEW intrusion then spread northward. The ocean response and 14dynamic of 16 process of the cold dome variability induced by Morakot have been detailed in several studies (i.e., [26]).

(a)

(b)

FigureFigure 10. 10.Daily Daily surface surface temperaturetemperature andand velocity at at surface surface (a (a) )deviating deviating from from the the day day before before the the TyphoonTyphoon Morakot Morakot (06 /08(06/08/2009)/2009) and and (b) deviating(b) deviating from thefrom day the before day thebefore Typhoon the Typhoon Haikui (05 Haikui/08/2012). Daily(05/08/2012). mean model Daily data mean were model provided data by were the HYCOM provided GLBa0.08 by the experiment.HYCOM GLBa0.08 The vectors experiment. are represented The byvectors the diff areerence represented of current by inthe zonal difference and meridional of current in components. zonal and meridional components.

Figure 10b shows the daily temperature and velocity anomaly field during Typhoon Haikui. During the forced period (August 7 to 8), ocean surface cooling occurred in the cold dome area. Due to the different relative locations of the TS and the typhoon, compared with the Morakot case, the flow in the TS did not turn southward in the forced period daily-mean pattern. The wind-induced strong eastward flow hindered the Kuroshio intrusion on the ECS continental shelf. Following the storm forced period, ocean surface warming occurred in the cold dome area. The SST rose even higher than under the prestorm conditions. Wind stress favored the eastward spreading of the Taiwan Current during Typhoon Haikui. The warming may have been associated with due to

Remote Sens. 2020, 12, 3321 13 of 14

Figure 10b shows the daily temperature and velocity anomaly field during Typhoon Haikui. During the forced period (August 7 to 8), ocean surface cooling occurred in the cold dome area. Due to the different relative locations of the TS and the typhoon, compared with the Morakot case, the flow in the TS did not turn southward in the forced period daily-mean pattern. The wind-induced strong eastward flow hindered the Kuroshio intrusion on the ECS continental shelf. Following the storm forced period, ocean surface warming occurred in the cold dome area. The SST rose even higher than under the prestorm conditions. Wind stress favored the eastward spreading of the Taiwan Current during Typhoon Haikui. The warming may have been associated with downwelling due to flow convergence driven by the strong wind. This downwelling reduced the baroclinity of the Kuroshio Current off the northeastern coast of Taiwan, spreading the warm surface water in the Kuroshio region toward the ECS continental shelf. As the typhoon moved through the ECS continental shelf, flow divergence occurred in the upper layer along with convergence between ECS and Kuroshio waters off northeastern Taiwan, as evident in the August 7 daily-mean pattern (Figure 10). When studying typhoon-induced ocean response, the daily-mean pattern might be insufficient for observing essential features with a time scale smaller than one day. Further investigation is needed to detail the dynamic process.

5. Conclusions In conclusion, the present study used multi-satellite observations, HYCOM model data, and weather station data to investigate the relationship between the intensity of cooling and wind stress, wind stress curl, TS northward transport, and Kuroshio intrusion. The SST response associated with a typhoon may be either cooling or warming [12]. The results suggest that average westward and northward wind stress are positively correlated with upwelling intensity. The decrease in TS northward transport created a condition that favored the Kuroshio intrusion. The decrease in TS transport was also positively correlated with the intensity of cooling. However, Ekman pumping associated with wind stress curl was less highly correlated with cooling strength.

Author Contributions: Y.-C.K. and M.-A.L. conceptualize and designed the method; Y.-C.K. implemented the method and performed the experiments; M.-A.L. and Y.C. discussed the results and revised the manuscript. All authors have read and agreed to the published version of the manuscript. Funding: This study was part of the Taiwan Integrated Research Program on Change Adaptation Technology (TaiCCAT), sponsored by grants from the Ministry of Science and Technology of Taiwan, MOST 104-2621-M-019-001 and 104-2811-M-019-001. The research was partly funded by the Center of Excellence for Oceans, National Taiwan Ocean University. Conflicts of Interest: The authors declare no conflict of interest.

References

1. Tang, T.Y.; Hsueh, Y.; Yang, Y.J.; Ma, J.C. Continental Slope Flow Northeast of Taiwan. J. Phys. Oceanogr. 1999, 29, 1353–1362. [CrossRef] 2. Jan, S.; Chen, C.-C.; Tsai, Y.-L.; Yang, Y.J.; Wang, J.; Chern, C.-S.; Gawarkiewicz, G.; Lien, R.-C.; Centurioni, L.; Kuo, J.-Y. Mean Structure and Variability of the Cold Dome Northeast of Taiwan. Oceanography 2011, 24, 100–109. [CrossRef] 3. Shen, M.-L.; Tseng, Y.; Jan, S. The formation and dynamics of the cold-dome off northeastern Taiwan. J. Mar. Syst. 2011, 86, 10–27. [CrossRef] 4. Chern, C.-S.; Wang, J.; Wang, D.-P. The exchange of Kuroshio and shelf water. J. Geophys. Res. Space Phys. 1990, 95, 16017–16023. [CrossRef] 5. Chao, S.-Y. Circulation of the East China Sea, a numerical study. J. Oceanogr. 1990, 46, 273–295. [CrossRef] 6. Hsueh, Y.; Wang, J.; Chern, C.-S. The intrusion of the Kuroshio across the continental shelf northeast of Taiwan. J. Geophys. Res. Space Phys. 1992, 97, 14323–14330. [CrossRef] 7. Chern, C.S.; Wang, J. The influence of Taiwan Strait waters on the circulation of the southern East China Sea. La Mer 1992, 30, 223–228. 8. Tang, T.; Tai, J.; Yang, Y.J. The flow pattern north of Taiwan and the migration of the Kuroshio. Cont. Shelf Res. 2000, 20, 349–371. [CrossRef] Remote Sens. 2020, 12, 3321 14 of 14

9. Wu, C.-R.; Lu, H.-F.; Chao, S.-Y. A numerical study on the formation of upwelling off northeast Taiwan. J. Geophys. Res. Space Phys. 2008, 113, C08025. [CrossRef] 10. Chang, Y.; Liao, H.T.; Lee, M.A.; Chan, J.W.; Shieh, W.J.; Lee, K.T.; Wang, G.H.; Lan, Y.C. Multisatellite observation on upwelling after the passing of Typhoon Hai-Tang in the southern East China Sea. Geophys. Res. Lett. 2008, 35, L03612. [CrossRef] 11. Siswanto, E.; Morimoto, A.; Kojima, S. Enhancement of phytoplankton primary in the southern East China Sea following episodic typhoon passage. Geophys. Res. Lett. 2009, 36, L11603. [CrossRef] 12. Tsai, Y.; Chern, C.-S.; Wang, J. The upper ocean response to a moving typhoon. J. Oceanogr. 2008, 64, 115–130. [CrossRef] 13. Price, J.F. Upper ocean response to a hurricane. J. Phys. Oceanogr. 1981, 11, 153–175. [CrossRef] 14. Tsai, Y.; Chern, C.-S.; Wang, J. Typhoon induced upper ocean cooling off northeastern Taiwan. Geophys. Res. Lett. 2008, 35, L14605. [CrossRef] 15. Morimoto, A.; Kojima, S.; Jan, S.; Takahashi, D. Movement of the Kuroshio axis to the northeast shelf of Taiwan during typhoon events. Estuar. Coast. Shelf Sci. 2009, 82, 547–552. [CrossRef] 16. Sakaida, F.; Kawamura, H. Accuracies of NOAA/NESDIS sea surface temperature estimation technique in the oceans around Japan. J. Oceanogr. 1992, 48, 345–351. [CrossRef] 17. McClain, E.P.; Pichel, W.G.; Walton, C.C. Comparative performance of AVHRR-based multichannel sea surface temperatures. J. Geophys. Res. Space Phys. 1985, 90, 11587–11601. [CrossRef] 18. Lee, M.-A.; Chang, Y.; Sakaida, F.; Kawamura, H.; Cheng, C.-H.; Chan, J.-W.; Huang, I. Validation of Satellite-Derived Sea Surface Temperatures for Waters around Taiwan. Terr. Atmos. Ocean. Sci. 2005, 16, 1189–1204. [CrossRef] 19. Large, W.G.; Pond, S. Open ocean momentum flux measurements in moderate to strong winds. J. Phys. Oceanogr. 1981, 11, 324–336. [CrossRef] 20. Guan, C.; Xie, L. On the Linear Parameterization of Drag Coefficient over Sea Surface. J. Phys. Oceanogr. 2004, 34, 2847–2851. [CrossRef] 21. Smith, S.D. Wind Stress and Heat Flux over the Ocean in Gale Force Winds. J. Phys. Oceanogr. 1980, 10, 709–726. [CrossRef] 22. Wu, J. Wind-Stress coefficients over Sea surface near Neutral Conditions—A Revisit. J. Phys. Oceanogr. 1980, 10, 727–740. [CrossRef] 23. Shih, Y.-Y.; Hsieh, J.-S.; Gong, G.-C.; Hung, C.-C.; Chou, W.-C.; Lee, M.-A.; Chen, K.-S.; Chen, M.-H.; Wu, C.-R. Field Observations of Changes in SST, Chlorophyll and POC Flux in the Southern East China Sea Before and After the Passage of Typhoon Jangmi. Terr. Atmos. Ocean. Sci. 2013, 24, 899. [CrossRef] 24. Price, J.F.; Sanford, T.B.; Forristall, G.Z. Forced Stage Response to a Moving Hurricane. J. Phys. Oceanogr. 1994, 24, 233–260. [CrossRef] 25. Tsai, Y.; Chern, C.-S.; Jan, S.; Wang, J. Numerical Study of Cold Dome Variability induced by Typhoon Morakot (2009) off Northeastern Taiwan. J. Mar. Res. 2013, 71, 109–131. [CrossRef] 26. Gawarkiewicz, G.; Jan, S.; Lermusiaux, P.; McClean, J.; Centurioni, L.; Taylor, K.; Cornuelle, B.; Duda, T.; Wang, J.; Yang, Y.J.; et al. Circulation and Intrusions Northeast of Taiwan: Chasing and Predicting Uncertainty in the Cold Dome. Oceanography 2011, 24, 110–121. [CrossRef]

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).